This project use feature extractor and classification algorithm to detect 0 - 11 digit of Hoda dataset.
- HODA Farsi Digit Dataset : http://farsiocr.ir
- KNN
- Bayes
- MLP
- RBF
- AdaBoost
- GSD
- Parzen
- LinearSVC
- HOG
- SVD
- PCA
This code is written in python. To use it you will need:
- python3
- matplotlib
- sklearn
- numpy
- cv2
Run python PHDRI.py
| Algorithm | Feature-Extrector | Accuracy |
|---|---|---|
| KNN | HOG | 91% |
| KNN | SVD | 94% |
| KNN | PCA | 85% |
| MLP | HOG | 93% |
| MLP | SVD | 57% |
| MLP | PCA | 90% |
| Parzen | HOG | 91% |
| Parzen | SVD | 70% |
| Parzen | PCA | 83.2% |
| Bayes | HOG | 91% |
| Bayes | SVD | 75% |
| Bayes | PCA | 85% |
| AdaBoost | HOG | 48% |
| AdaBoost | SVD | 50% |
| AdaBoost | PCA | 30% |
| LinearSVC | HOG | 85% |
| LinearSVC | SVD | 70% |
| LinearSVC | PCA | 60% |
| RBF | HOG | 88% |
| RBF | SVD | 83% |
| RBF | PCA | 80% |